A high-resolution time-frequency analysis tool for fault diagnosis of rotating machinery

Fault features in mechanical systems often manifest as transient impulses, which can be effectively analyzed using time-frequency analysis (TFA) methods. Recently, a new TFA technique known as the time-reassigned multisynchrosqueezing transform (TMSST) was proposed to capture these transient impulses for fault diagnosis. However, the TMSST, which is based on the short-time Fourier transform (STFT), suffers from unclear high-frequency representations owing to the fixed sliding window used in the STFT. To address this limitation, the current study combined TMSST with the S-transform and a local maximum method to enhance the time-frequency representation for improved signal analysis. Furthermore, an extractive reconstruction algorithm that binds the maximum value of the spectral envelope is proposed for spectral decomposition. To validate the proposed technique, a simulated noise-added signal and four experimental bearing defect datasets were used. The results demonstrate that the proposed technique can effectively and accurately extract fault features from bearing signals regardless of whether the bearings operate under constant or varying speed conditions. This study offers a novel and efficient approach for fault diagnosis in mechanical systems with complex dynamic behaviors.